Bayesian evaluation of budgets for endemic disease control: An example using management changes to reduce milk somatic cell count early in the first lactation of Irish dairy cows

The aim of this research was to determine budgets for specific management interventions to control heifer mastitis in Irish dairy herds as an example of evidence synthesis and 1-step Bayesian micro-simulation in a veterinary context. Budgets were determined for different decision makers based on their willingness to pay. Reducing the prevalence of heifers with a high milk somatic cell count (SCC) early in the first lactation could be achieved through herd level management interventions for pre- and peri-partum heifers, however the cost effectiveness of these interventions is unknown. A synthesis of multiple sources of evidence, accounting for variability and uncertainty in the available data is invaluable to inform decision makers around likely economic outcomes of investing in disease control measures. One analytical approach to this is Bayesian micro-simulation, where the trajectory of different individuals undergoing specific interventions is simulated. The classic micro-simulation framework was extended to encompass synthesis of evidence from 2 separate statistical models and previous research, with the outcome for an individual cow or herd assessed in terms of changes in lifetime milk yield, disposal risk, and likely financial returns conditional on the interventions being simultaneously applied. The 3 interventions tested were storage of bedding inside, decreasing transition yard stocking density, and spreading of bedding evenly in the calving area. Budgets for the interventions were determined based on the minimum expected return on investment, and the probability of the desired outcome. Budgets for interventions to control heifer mastitis were highly dependent on the decision maker's willingness to pay, and hence minimum expected return on investment. Understanding the requirements of decision makers and their rational spending limits would be useful for the development of specific interventions for particular farms to control heifer mastitis, and other endemic diseases.     

Constructing naive Bayesian classifiers for veterinary medicine: A case study in the clinical diagnosis of classical swine fever

For diseases of which the clinical diagnosis is uncertain, naive Bayesian classifiers can be of assistance to the veterinary practitioner. These simple probabilistic models have proven to be very powerful for solving classification problems in a variety of domains, but are not yet widely applied within the veterinary domain. In this paper, naive Bayesian classifiers and methods for their construction are reviewed. We demonstrate how to construct full and selective classifiers from a data set and how to build such classifiers from information in the literature. As a case study, naive Bayesian classifiers to discriminate between classical swine fever (CSF)-infected and non-infected pig herds were constructed from data collected during the 1997/1998 CSF epidemic in the Netherlands. The resulting classifiers were studied in terms of their accuracy and compared with the optimally efficient diagnostic rule that was reported earlier by Elbers et al. (2002). The classifiers were found to have accuracies within the range of 67-70% and performed comparable to or even better than the diagnostic rule on the available data. In contrast with the diagnostic rule, the classifiers had the advantage of taking both the presence and the absence of particular clinical signs into account, which resulted in more discriminative power. These results indicate that naive Bayesian classifiers are promising tools for solving diagnostic problems in the veterinary field.     

Bayesian estimates of genetic parameters for metritis, retained placenta, milk fever, and clinical mastitis in Holstein dairy cows via Gibbs sampling

The objective of this study was to estimate heritability and genetic correlations between the liabilities of clinical mastitis (CM), milk fever (MF), metritis (MET), and retained placenta (RP) within the first three lactations of Holstein dairy cows. The records of 57,301 dairy cows from 20 large dairy herds in Iran between January 2005 and June 2009 were analysed with univariate and bivariate threshold animal models, using Gibbs sampling methodology. The final model included the fixed class effects of herd-year, season of calving, parity of dam, the linear covariate effect of age at calving, and the random direct genetic effect of animal. Posterior means of heritability for liabilities in first, second, and third lactations were 0.06, 0.08, and 0.09, respectively, for CM; 0.10, 0.12, and 0.11, respectively, for MF; 0.09, 0.07, and 0.10, respectively, for MET, and 0.07, 0.08, and 0.08, respectively, for RP. Posterior means of genetic correlations between disease liabilities were low or moderate (from −0.01 to 0.26). The results of this study indicated the importance of health traits for considering in the selection index of Iranian Holstein dairy cows.     

Effects of a mastitis control programme on the incidence of clinical mastitis

Summary

The results of a mastitis control field experiment in Utrecht over three and a half years, including sevens herds with a total of 225 lactating cows are presented. Every case of clinical mastitis was examined bacteriologically. Quarter samples were taken routinely at an average interval of five weeks.

The incidence of clinical mastitis in each herd at the start of the experiment varied from 10 to 104 quarter cases per 100 cows per year. The majority of cases of clinical mastitis (33.1 per cent) occurred during the first month of lactation.

In 74.4 per cent of the lactations marked by clinical mastitis only a single case was observed during that lactation period. If the animals were on pasture during the first month of lactation, the incidence of clinical mastitis was significantly lower during this month, compared with the first month when they were housed. Str. dysgalactiae was the most common isolate from clinical cases (21.5 per cent). A total proportion of 40.3 per cent of the clinical cases were caused by cocci, 20.3 per cent by Gram‐negative bacteria and 16.0 per cent were bacteriologically negative. Of the clinical cases caused by streptococci and staphylococci, 33.3 per cent were preceded by subclinical infection, compared with 11.8 per cent of the clinical cases due to E. coli. The overall incidence of clinical mastitis in this experiment decreased from 47 to 31 per 100 cows per year.     

A Bayesian approach to the accuracy of clinical observations

The accuracy of clinical observations was estimated using Bayesian latent-class models with two or more independent tests. Four veterinarians carried out systematic independent clinical examinations on 155 pigs in three herds. Based on the results of binary recordings of clinical observations on dullness, poor body condition (PBC), skin lesions, lameness, respiratory disease, and diarrhea, a latent disease state for each clinical disease was estimated using Gibbs sampling.

The accuracy of the clinical observations differed for the four observers and for different clinical signs. Population parameters were estimated from a Bayesian hierarchical model, and the accuracy of a random observer was calculated. We concluded that the accuracy of the veterinarians in this study substantiated the need to pursue more-precise definitions of the clinical findings and that larger sample sizes would be needed to provide reasonable variance estimates. Finally, we concluded that the uncertainty in the clinical decision-making process (starting with the clinical examination) needs to be represented fully.     

Association analysis of the constructed linkage maps covering TLR2 and TLR4 with clinical mastitis in Norwegian Red cattle.

Toll-like receptors (TLR) are important cell-surface molecules mediating immune responses. Previous studies have identified TLR2 and TLR4 as potential candidate genes for disease resistance. In this study, dense linkage maps comprising single nucleotide polymorphisms (SNPs) have been constructed for the chromosomal regions harbouring TLR2 and TLR4 on bovine chromosome 17 and 8. The most likely marker orders for both regions were compared with the corresponding human map positions and used to reorder bovine scaffolds available from the bovine genome sequence assembly (Btau_3.1). A combined linkage and linkage disequilibrium method was used to investigate possible associations between the TLR genes and mastitis susceptibility recorded in the Norwegian Red cattle population. The analysis did not detect any significant association between the chromosomal regions surrounding TLR2 and TLR4 and mastitis in Norwegian Red cattle.     

A field study evaluation of Petrifilm™ plates as a 24-h rapid diagnostic test for clinical mastitis on a dairy farm

Clinical mastitis is one of the most common and expensive diseases of dairy cattle. To make an informed treatment decision, it is important to know the causative pathogen. However, no detection of bacterial growth can be made in approximately 30% of all clinical cases of mastitis. Before selecting the treatment regimen, it is important to know whether the mastitis-causing pathogen (MCP) is Gram-positive or Gram-negative. The aim of this field study was to investigate whether using two 3M Petrifilm™ products on-farm (which conveys a higher degree of sample freshness but also bears a higher risk for contamination than working in a lab) as 24-h rapid diagnostic of clinical mastitis achieved results that were comparable to the conventional microbiological diagnostic method.     

Investigation of the association between the interval from internal teat sealant administration to calving and early lactation clinical mastitis in New Zealand dairy heifers

AIMS: To examine the association between the interval from internal teat sealant (ITS) administration to calving and the incidence of farmer-recorded clinical mastitis in the first 30 days of lactation in pasture-based dairy heifers.

METHODS: Heifers that were administered an ITS by a single veterinary business in the South Island of New Zealand over the winter of 2014 were enrolled in a cross-sectional observational study. ITS was administered to all heifers on each participating farm on a single calendar day. The dates of calving and farmer-diagnosed clinical mastitis were recorded by farm staff. The interval from ITS administration to calving was categorised into four approximately evenly sized groups: <35, 35–48, 49–69 and >69 days. The quartile of the farm’s calving period in which each heifer calved was also investigated as a potential confounding variable. A hierarchical logistic regression model was constructed to determine the association between the interval from ITS administration to calving with the odds of clinical mastitis in the first 30 days of lactation.

RESULTS: Analysis was performed on 7,126 eligible heifers from 31 farms, with ITS administered between 9 May and 11 July 2014. The mean interval from ITS administration to calving was 52.9 (SD 24.4, min 1, max 137) days. Clinical mastitis was diagnosed in 420/7,126 (5.9 (95% CI=5.4–6.5)%) heifers between calving and day 30 of lactation. In the final multivariable model, which included calving period quartile, interval from ITS administration to calving was not associated with the odds of clinical mastitis (p=0.516). Compared to an interval from ITS administration to calving of <35 days, the adjusted OR of clinical mastitis for intervals of 35–48, 49–69 and >69 days were 0.83 (95% CI=0.59–1.17), 0.71 (95% CI=0.45–1.11) and 0.68 (95% CI=0.36–1.29), respectively.

CONCLUSIONS AND CLINICAL RELEVANCE: Within the range of intervals from ITS administration to calving observed in this study, there was no association with the odds of clinical mastitis in the first 30 days of lactation in dairy heifers. This study suggests that veterinary clinics may be able to extend their ITS administration service and treat dairy heifers earlier than the current recommendation of approximately 4 weeks before the planned start of calving.     

Antimicrobial usage and risk of retreatment for mild to moderate clinical mastitis cases on dairy farms following on-farm bacterial culture and selective therapy

AIMS: To assess antimicrobial usage for treatment of mild to moderate clinical mastitis, and risk of retreatment, following implementation of an on-farm bacterial culture system and selective therapy based on culture results, and to assess compliance with treatment decision tree protocols and the level of agreement between results from on-farm culture and laboratory-based microbiology methods.

MATERIALS AND METHODS: Herdowners from seven dairy herds were asked to collect milk samples from cases of mild to moderate clinical mastitis between July 2015 and May 2016. All samples were cultured on-farm using a commercially available selective media and were also submitted for laboratory-based culture. Within sequential pairs of cows with mastitis, half were assigned to be treated without regard to culture results (Blanket group), and half were treated based on the on-farm culture results (Selective group) according to decision tree diagrams provided to the farmers. Culture results, treatments, and retreatments for clinical mastitis were recorded. The sum of the daily doses of antimicrobials used per cow, the number of retreatments and interval to first retreatment were compared between treatment groups.

RESULTS: The geometric mean sum of daily doses for quarters assigned to the Selective (1.72 (95% CI=1.55–1.90)) group was lower than for the Blanket (2.38 (95% CI=2.17–2.60)) group (p=0.005). The percentage of cows retreated for clinical mastitis did not differ between the Selective (21.7 (95% CI=10.5–25.9)%) and Blanket (26.1 (95% CI=20.9–31.3)%) groups (p=0.13), and there was no difference between groups in the hazard that cows would be retreated within 60 days of enrolment (hazard ratio=0.82 (95% CI=0.39–1.69); p=0.59). Compliance with the treatment protocols was higher amongst quarters assigned to the Selective (199/233; 85.4%) compared with the Blanket (171/249; 68.7%) group (p<0.001), and varied between farms from 64–94%. The overall agreement between results from on-farm and laboratory culture was 188/331 (56.9%; kappa=0.31; p<0.001), but varied between farms from 44.7–88.2% (p<0.001).

CONCLUSIONS: Use of on-farm culture with selective antimicrobial therapy resulted in approximately 25% lower antimicrobial usage, but was not associated with an increase in the proportion of cows retreated for clinical mastitis.

CLINICAL RELEVANCE: This study has demonstrated that on-farm culture and selective therapy based on culture results can be implemented on-farm. However, farms varied in their implementation of both the treatment protocols and microbiology procedures. Where such systems are to be used on-farm, specific training and on-going monitoring is required.     

Considerations for analysis of time-to-event outcomes subject to competing risks in veterinary clinical studies

In veterinary medicine, prospective clinical trials are increasingly utilized to address questions regarding effectiveness of therapies and patient prognosis. A large number of these trials involve time-to-event (TTE) endpoints, which require special methods of analysis to handle data in which not all subjects are observed to have the event of interest. Analyses and interpretation of the results can be further complicated when an endpoint of interest is not observed in some patients because they incur a competing risk, such as death from an unrelated cause. Competing risks have been the source of confusion in many epidemiologic analyses leading to the potential for misinterpretation. In this article, we review key considerations for the TTE analysis in the setting of competing risks. We briefly review standard TTE tools, namely Kaplan–Meier survival curves and Cox regression. In the setting of outcomes with competing risks, we provide guidance on the appropriate analysis techniques, such as cumulative incidence curves, to estimate the risk of an event of interest. We also describe a common pitfall of treating competing risks as censoring in Kaplan–Meier survival curve analysis, which can overestimate the event rate of interest. We describe two common regression methods that examine associated risk factors in the presence of competing risks and highlight the different research questions these methods address. This article provides an introductory overview and illustrates concepts with examples from veterinary trials and with example data sets.     

Building a picture: Prioritisation of exotic diseases for the pig industry in Australia using multi-criteria decision analysis

Diseases that are exotic to the pig industry in Australia were prioritised using a multi-criteria decision analysis framework that incorporated weights of importance for a range of criteria important to industry stakeholders. Measurements were collected for each disease for nine criteria that described potential disease impacts. A total score was calculated for each disease using a weighted sum value function that aggregated the nine disease criterion measurements and weights of importance for the criteria that were previously elicited from two groups of industry stakeholders. One stakeholder group placed most value on the impacts of disease on livestock, and one group placed more value on the zoonotic impacts of diseases. Prioritisation lists ordered by disease score were produced for both of these groups. Vesicular diseases were found to have the highest priority for the group valuing disease impacts on livestock, followed by acute forms of African and classical swine fever, then highly pathogenic porcine reproductive and respiratory syndrome. The group who valued zoonotic disease impacts prioritised rabies, followed by Japanese encephalitis, Eastern equine encephalitis and Nipah virus, interspersed with vesicular diseases. The multi-criteria framework used in this study systematically prioritised diseases using a multi-attribute theory based technique that provided transparency and repeatability in the process. Flexibility of the framework was demonstrated by aggregating the criterion weights from more than one stakeholder group with the disease measurements for the criteria. This technique allowed industry stakeholders to be active in resource allocation for their industry without the need to be disease experts. We believe it is the first prioritisation of livestock diseases using values provided by industry stakeholders. The prioritisation lists will be used by industry stakeholders to identify diseases for further risk analysis and disease spread modelling to understand biosecurity risks to this industry.     

Objective interpretation of bovine clinical biochemistry data: application of Bayes law to a database model

With the advent of animal-side biochemistry analysers in veterinary practice, the requirement for ready access to reliable means for interpretation of the results is of increasing importance. At the University of Glasgow Veterinary School (GUVS), a large computerised hospital database containing extensive clinical, laboratory, and pathological information has been maintained. A retrospective study was undertaken to investigate plasma biochemistry results and corresponding post mortem diagnosis data from 754 unwell cattle which had presented to GUVS over the study period. Initial analysis of the clinical biochemistry data from this unwell population revealed that the parameters did not follow a normal distribution. This finding suggested that the accepted reference range method for the interpretation of clinical biochemistry data may provide limited information about the unwell animal. By applying a combination of percentile analysis and conditional probability techniques to the hospital data, the development of a means of clinical biochemistry interpretation was developed whereby a clinician could determine whether a value was abnormal, the degree of abnormality, and the most likely associated diseases. For example, a urea value of 30 mmol/1 lay within the top 5% of results, and one of the most common diseases associated with this urea value was pyelonephritis. Furthermore, a Bayesian approach allowed the quantification of the relationship between any plasma biochemistry value and disease through the generation of a ratio termed the ‘biochemical factor’. Using the same example, given a urea value of 30 mmol/1, pyelonephritis was eight times more likely than before any biochemistry information was known. The results from the study were used to form the basis of a software system which may ultimately be used by the clinician to aid in the recognition, treatment and prevention of disease in the veterinary domain.     

兽用B超在母猪早期妊娠监测中的临床应用及效益分析

超声波扫描技术具有无损伤、无放射性危害、诊断准确率高、重复性好、安全无副作用、易操作等特点。本文探讨了超声波扫描在母猪早期妊娠监测中的应用、探测方法、诊断标准及效益分析。结果表明：超声探查技术能够直接对妊娠动物的妊娠诊断、胎龄判定、胎儿活力、胎儿数目及胚胎早期死亡或流产做出直观、准确的诊断,相对传统诊断技术具有明显优势。     

The relative effect of genomic information on efficiency of Bayesian analysis of the mixed linear model with unknown variance

This work focuses on the effects of variable amount of genomic information in the Bayesian estimation of unknown variance components associated with single‐step genomic prediction. We propose a quantitative criterion for the amount of genomic information included in the model and use it to study the relative effect of genomic data on efficiency of sampling from the posterior distribution of parameters of the single‐step model when conducting a Bayesian analysis with estimating unknown variances. The rate of change of estimated variances was dependent on the amount of genomic information involved in the analysis, but did not depend on the Gibbs updating schemes applied for sampling realizations of the posterior distribution. Simulation revealed a gradual deterioration of convergence rates for the locations parameters when new genomic data were gradually added into the analysis. In contrast, the convergence of variance components showed continuous improvement under the same conditions. The sampling efficiency increased proportionally to the amount of genomic information. In addition, an optimal amount of genomic information in variance–covariance matrix that guaranty the most (computationally) efficient analysis was found to correspond a proportion of animals genotyped ***0.8. The proposed criterion yield a characterization of expected performance of the Gibbs sampler if the analysis is subject to adjustment of the amount of genomic data and can be used to guide researchers on how large a proportion of animals should be genotyped in order to attain an efficient analysis.     

The use of Markov chain Monte Carlo for analysis of correlated binary data: patterns of somatic cells in milk and the risk of clinical mastitis in dairy cows

Two analytical approaches were used to investigate the relationship between somatic cell concentrations in monthly quarter milk samples and subsequent, naturally occurring clinical mastitis in three dairy herds. Firstly, cows with clinical mastitis were selected and a conventional matched analysis was used to compare affected and unaffected quarters of the same cow. The second analysis included all cows, and in order to overcome potential bias associated with the correlation structure, a hierarchical Bayesian generalised linear mixed model was specified. A Markov chain Monte Carlo (MCMC) approach, that is Gibbs sampling, was used to estimate parameters.

The results of both the matched analysis and the hierarchical modelling suggested that quarters with a somatic cell count (SCC) in the range 41,000–100,000 cells/ml had a lower risk of clinical mastitis during the next month than quarters <41,000 cell/ml. Quarters with an SCC >200,000 cells/ml were at the greatest risk of clinical mastitis in the next month. There was a reduced risk of clinical mastitis between 1 and 2 months later in quarters with an SCC of 81,000–150,000 cells/ml compared with quarters below this level. The hierarchical modelling analysis identified a further reduced risk of clinical mastitis between 2 and 3 months later in quarters with an SCC 61,000–150,000 cells/ml, compared to other quarters.

We conclude that low concentrations of somatic cells in milk are associated with increased risk of clinical mastitis, and that high concentrations are indicative of pre-existing immunological mobilisation against infection. The variation in risk between quarters of affected cows suggests that local quarter immunological events, rather than solely whole cow factors, have an important influence on the risk of clinical mastitis. MCMC proved a useful tool for estimating parameters in a hierarchical Bernoulli model. Model construction and an approach to assessing goodness of model fit are described.     

Predicting the probability of abortion in dairy cows: a hierarchical Bayesian logistic-survival model using sequential pregnancy data

Although abortion contributes substantially to poor reproductive health of dairy herds, little is known about the predictability of abortion based on age, previous abortion or gravidity (number of previous pregnancies). A poor understanding of effects of maternal factors on abortion risk exists, in part, because of methodological difficulties related to non-independence of multiple pregnancies of the same cow in analysis of fetal survival data. We prospectively examined sequential pregnancies to investigate relationships between fetal survival and putative dam risk factors for 2991 abortions from 24,706 pregnancies of 13,145 cows in nine California dairy herds. Relative risks and predicted probabilities of abortion (PPA) were estimated using a previously described hierarchical Bayesian logistic-survival model generalized to incorporate longitudinal data of multiple pregnancies from a single cow. The PPA increased with increasing dam age at conception, with increasing number of previous abortions, and if the previous pregnancy was aborted >60 days in gestation. The PPA decreased with increasing gravidity and with increasing number of days open. For cows that aborted, the median time to fetal death decreased slightly as gravidity increased. The study considers several methodological issues faced in epidemiologic investigations of fetal health, including multi-modal hazard functions, extensive censoring and non-independence of multiple pregnancies. The model improves our ability to predict bovine abortion and to characterize fetal survival, which have important applications to herd health management.     

Genetic analysis of liability to clinical mastitis, with somatic cell score and production traits using bivariate threshold–linear and linear–linear models

Inferences about genetic and residual correlation estimates and sire evaluations involving a categorical trait with linear model are ambiguous and mostly based on data simulations. In this study, estimates of variance components and prediction of breeding values in a model with a categorical and a continuous trait were compared between threshold–linear (TLM) and linear–linear models (LLM) in analysis of large clinical mastitis (CM) field data. Data on CM, somatic cell score (SCS), 305-day milk (MY), protein (PY) and fat yield (FY) from first-lactation Finnish Ayrshire cows were used. Four bivariate analyses were made using a TLM in Bayesian analysis. Each analysis fitted CM and one continuous trait at a time. Corresponding bivariate analyses were made using a Gaussian linear model. Estimates of heritabilities for CM were 0.06 and 0.02 from TLM and LLM, respectively whilst heritability estimates of the continuous traits were similar from both models. Genetic correlations between CM–SCS, CM–MY, CM–PY, and CM–FY from TLM and LLM were 0.63 and 0.63; 0.36 and 0.36; 0.32 and 0.32; 0.30 and 0.29, respectively. Estimates of residual correlations were 0.11 and 0.06; − 0.04 and − 0.02; − 0.03 and − 0.02; − 0.05 and − 0.03 between CM–SCS, CM–MY, CM–PY, and CM–FY, respectively. Comparison between the models indicates similar estimates of genetic correlations with no underestimation with the linear model analysis. In CM evaluation, the comparison of model's predictive ability showed greater improvements in accuracy with the bivariate than with the univariate models. There was, however no clear advantage of univariate threshold model over univariate linear model, except for less accuracy sires.     

Investigations on the influence of serum D-lactate levels on clinical signs in calves with metabolic acidosis

Correlations between the degree of acidosis and clinical signs (changes in posture, behaviour, intensity of suckling reflex) in neonatal diarrhoeic calves have been described in various studies. However, base excess values varied widely in calves exhibiting similar clinical symptoms. The objective of this study was to elucidate whether the clinical picture of acidotic calves with neonatal diarrhoea is influenced more by D-lactate concentration than by degree of acidosis. Eighty calves up to three weeks old that were admitted to the II Medical Animal Clinic with acute diarrhoea and base excess values between -10 and -25 mmol/L were included in the prospective study. Posture, behaviour, suckling and palpebral reflexes, and position of the eyeballs were scored during the initial examination. Base excess and serum D-lactate and urea concentrations were determined in venous blood. In order to quantify the influences of base excess and d-lactate on the clinical parameters, groups of different clinical categories were compared. The results show that variations in behaviour, and in posture can be better explained by elevations of serum D-lactate concentrations than by decreases in base excess. Disturbances of the palpebral reflex appear to be almost completely caused by high levels of D-lactate.     

A benefit-cost analysis of veterinary interventions in Afghanistan based on a livestock mortality study

This article evaluates a veterinary intervention program of the Dutch Committee for Afghanistan, started during the Soviet invasion of Afghanistan, a 10 year period during which veterinary services were otherwise completed disrupted. The veterinary field program was carried out mainly by paravets. A questionnaire-based survey was conducted to measure livestock mortality in districts that received veterinary services through the project and in neighboring districts that had not had veterinary services for about 10 years. The survey indicated that livestock mortality in districts that received veterinary services was lower than in districts without any veterinary services. Comparison of each of the 22 age-species-specific district pairs showed a difference in favor of the covered district in 18 pairs. In 12 out of these 18 pairs, this difference was significant. Overall annual mortality rates differed (in relative amounts) by 25%, 30%, and 22%, in calves, lambs, and kids, respectively, and in adult cattle, sheep, and goats, by roughly 30%, 40%, and 60%, respectively. In the absence of any other obvious distinctive features between the compared districts, we concluded that this difference in mortality was due to the presence or absence of veterinary services. A benefit-cost analysis showed that the benefit-cost ratio for the program was between 1.8 and 4.8. The high benefit-cost ratio resulted partly from the fact that the costs of the program were low, because it was implemented by a volunteer-run, non-governmental organization. In addition, due to the special circumstances prevailing in the country, any input in combatting diseases at this stage was bound to have a relatively high impact. We concluded that: (1) the veterinary program was important for the rebuilding of numbers of livestock and thus for the economy of Afghanistan, and (2) veterinary intervention programs under these and comparable circumstances can be highly cost-effective.